MRI volumes of amygdala and hippocampus in non–mentally retarded autistic adolescents and adults
Citation Manager Formats
Make Comment
See Comments

Abstract
Objective: To determine whether volumes of hippocampus and amygdala are abnormal in people with autism.
Background: Neuropathologic studies of the limbic system in autism have found decreased neuronal size, increased neuronal packing density, and decreased complexity of dendritic arbors in hippocampus, amygdala, and other limbic structures. These findings are suggestive of a developmental curtailment in the maturation of the neurons and neuropil.
Methods: Measurement of hippocampus, amygdala, and total brain volumes from 1.5-mm coronal, spoiled gradient-recalled echo MRI scans in 14 non–mentally retarded autistic male adolescents and young adults and 14 individually matched, healthy community volunteers.
Results: Amygdala volume was significantly smaller in the autistic subjects, both with (p = 0.006) and without (p = 0.01) correcting for total brain volume. Total brain volume and absolute hippocampal volume did not differ significantly between groups, but hippocampal volume, when corrected for total brain volume, was significantly reduced (p = 0.04) in the autistic subjects.
Conclusions: There is a reduction in the volume of amygdala and hippocampus in people with autism, particularly in relation to total brain volume. The histopathology of autism suggests that these volume reductions are related to a reduction in dendritic tree and neuropil development, and likely reflect the underdevelopment of the neural connections of limbic structures with other parts of the brain, particularly cerebral cortex.
Autism is a behavioral syndrome defined by specific qualitative impairments in reciprocal social interaction and communication and symbolic play, and restricted and stereotyped patterns of behavior, interests, and activity.1 Neuropathologic and neuroimaging studies have provided evidence in autism of structural abnormalities of developmental origin in the cerebral cortex, portions of the limbic system, cerebellum, and inferior olivary nuclei.2,3 The findings suggest abnormalities in neuronal alignment and in the elaboration, pruning, and selective elimination of neuronal processes in cerebral cortex, the elaboration of dendritic and axonal ramifications in the limbic system, and the selective elimination of neurons in the cerebellum. Studies of brain function suggest decreased functional connections within cerebral cortex and between cerebral cortex and subcortical regions, and delayed maturation and dysfunction of frontal circuitry in autism.4-7 Collectively, these structural and functional findings have led to a growing recognition that the pathophysiology of autism involves the underdevelopment of the circuitry of neural systems that involve cerebral cortex, limbic system, and cerebellum.8 The neural systems involved have not yet been characterized, nor has the nature of the functional contribution of the individual structures to the involved neural systems.
Although not consistently supported,2 Bauman and Kemper3,9,10 have reported a pattern of small, immature-appearing neurons and increased neuronal packing density in the hippocampus and amygdala, as well as in other related subcortical structures, based on examination of serial whole-brain sections in nine well documented cases of autism. Golgi studies of CA1 and CA4 pyramidal cells revealed reduced complexity and extent of dendritic arbors, suggestive of truncation in the maturation of the neurons, their dendritic arbors, and the neuropil.11 Despite these neuropathologic findings, few attempts have been made in autism to measure these complex structures with neuroimaging. Only two MRI studies of the hippocampus in autism and none of amygdala have been published. Saitoh et al.,12 using cross-sectional areas on three 5-mm coronal slices, reported no differences in hippocampal volume between 33 autistic subjects (age range, 6 to 42 years) and 23 age-matched control subjects. Piven et al.13 assessed hippocampal volume on 1.5-mm images in 35 autistic and 36 control subjects, and found no group differences.
We studied MRI volumetric measurements of the hippocampus and amygdala and total brain volume in 14 individually matched pairs of non–mentally retarded autistic subjects and control subjects. We matched autistic and control subjects on age, race, and IQ, and studied only male subjects to control variables known to affect brain morphometry. We studied only people with IQ scores in the normal range to eliminate the nonspecific influence of mental retardation on structure size.
Methods.
Subjects.
Subjects were 14 individually matched pairs of male, non–mentally retarded autistic and normal people between 11 and 37 years of age (table 1). The autistic subjects represented all consecutive referrals to a research clinic who met inclusion and exclusion criteria for the study. Control subjects were community volunteers recruited as individual matches for the autistic subjects based on age, IQ, sex, and race; family of origin socioeconomic status was matched at a group level. All subjects were white. Inclusion criteria included Full Scale and Verbal IQ scores of 80 or above and sufficient cooperation to complete neuropsychologic testing and imaging without sedation. Methodology of the study, including MRI scanning of minors, was approved by the University of Pittsburgh Medical Center Institutional Review Board. Procedures were fully explained to all subjects and, when appropriate, to their parent or guardian. Written informed consent was obtained from subjects or their guardians.
Demographic characteristics of the individually matched pairs*
The diagnosis of autism was established through expert clinical evaluation in accordance with accepted clinical descriptions of high-functioning autistic people14 and two structured research diagnostic instruments, the Autism Diagnostic Interview15,16 and the Autism Diagnostic Observation Schedule.17 Audiotapes and videotapes of the administration of these two diagnostic instruments were reviewed by the developers of the instruments, thus providing independent confirmation of diagnosis in all cases. Potential autistic subjects were excluded if found to have evidence of an associated infectious, genetic, or metabolic disorder, such as fragile X syndrome, tuberous sclerosis, or fetal cytomegalovirus infection. Potential subjects were also excluded if found to have evidence of birth asphyxia, head injury, or a seizure disorder. Exclusions were based on neurologic history and examination, physical examination, chromosomal analysis, and, if indicated, metabolic testing.
Control subjects were medically healthy adolescents and young adults recruited from the community through advertisements in areas socioeconomically comparable with those of the families of origin of the autistic subjects. Potential control subjects were screened by questionnaire, telephone, face-to-face interview, and observation during screening psychometric tests. Exclusionary criteria, evaluated through these procedures, included current or past history of psychiatric and neurologic disorders, birth injury, developmental delay, school problems, acquired brain injury, learning disabilities, and medical disorders with implications for the CNS or requiring regular medication usage. Potential control subjects were also screened to exclude those with a family history of autism, developmental cognitive disorder, learning disability, affective disorder, anxiety disorder, schizophrenia, obsessive-compulsive disorder, or other neurologic or psychiatric disorders thought to have a genetic component. Control subjects were selected as individual matches for the autistic subjects based on age (within 6 months for those younger than 17 years and within 1 year for those older than 17 years), sex, Full Scale IQ (within 5 points), and race. Subjects were group matched on socioeconomic status of the family of origin as measured by the Hollingshead method.
MRI scans.
All scans were obtained on a General Electric (Milwaukee, WI) 1.5-T Signa scanner, using a standard protocol identical to that described in previous studies by our group (e.g., Aylward et al.18). This protocol includes a 1.5-mm SPGR (spoiled gradient-recalled echo in steady state) coronal series (TR = 35; TE = 5, number of excitations = 1, flip angle = 45 degrees), which was used for all measurements reported in this study. All images were transferred from the acquisition facility to the image analysis laboratory by File Transfer Protocol (FTP) and archived on CD-ROM disks. Scans were reconstructed into a three-dimensional image, and brains were reoriented, if necessary, to ensure that the coronal slices were perpendicular to the line connecting the anterior and posterior commissures (AC-PC line).
Measurement of total brain volume.
All measurements were made on a Gateway 2000 graphics workstation (North Sioux City, SD), using locally developed custom graphics software. A semiautomated thresholding procedure was used for segmenting brain from CSF and extracerebral tissue, as described elsewhere.18 Intrarater reliability for obtaining brain volumes with this procedure yielded an intraclass correlation of 0.99 on 10 brains.
Measurement of hippocampus and amygdala.
The rules used for defining the boundaries of the amygdala and hippocampus are described by Honeycutt et al.19 and on the Internet web site of the Johns Hopkins University Division of Psychiatric Neuroimaging (http://pni.med.jhu.edu). Boundaries of the hippocampus were traced, starting in the most posterior slice in which any hippocampal tissue could be observed. The choroid fissure served as the superior boundary, the inferior temporal horn of the lateral ventricle as the lateral boundary, and the white matter of the parahippocampal gyrus as the inferior boundary. The hippocampus forms a natural boundary around the edge of the mesial temporal lobe. Both the alveus and the subiculum were included in hippocampal measurements. Anteriorly, when a clear demarcation between the hippocampus and amygdala was not seen in the coronal view, the sagittal view was consulted to determine the border between the hippocampus and amygdala. The amygdala was measured in reconstructed axial slices (0.9375 mm thick, parallel to the AC-PC line) with the superior boundary set at the level of the tubera where the mamillary bodies and optic nerves can be clearly viewed. On each axial slice, a straight line was drawn from the most medial white matter protruding into the amygdaloid gray matter on each side to the lateral fissure; gray matter medial to this line was included in the measurement. The medial boundary was set at the uncus and the posterior boundary was set at the temporal horn of the lateral ventricle in superior slices and at the hippocampus in inferior slices. No gray matter inferior to the level of the hippocampus was included. After measuring in axial slices, the rater consulted the coronal views, and any tissue anterior to the anterior commissure was erased, as well as any tissue medial to the uncal notch. Area of each structure was calculated in each slice, and areas were summed across slices and multiplied by slice thickness, resulting in approximate volumes. Interrater reliability for the hippocampus (intraclass correlation = 0.97) and amygdala (intraclass correlation = 0.88) was established between two raters. Measurements were completed by one rater, who was blind to group status (autism versus control).
Statistical analysis.
Volumes of hippocampus and amygdala from autistic subjects were compared with volumes from the matched control subjects using paired t-tests. Structure volumes were considered in relation to whole-brain volume through ratio measures (structure volume divided by brain volume). Paired t-tests were also used to compare the autistic and control subjects on these ratio measures. Paired t-tests using ratio measures were chosen as the most appropriate method for examining the specificity of regional volume differences, because this method (unlike analysis of covariance) takes advantage of the extra power afforded by the careful individual matching. An analysis of covariance was used to compare autistic and control subjects on structure volumes, controlling for height, although this resulted in reduction of power with loss of the matched-pair design. A repeated-measures analysis of variance was used to assess group differences in asymmetries of hippocampus and amygdala. Right and left structure volumes were the repeated measures, and diagnosis (autism versus control) was the between-subjects variable. All reported p values are for two-tailed tests.
Results.
There were no significant differences between the autistic subjects and individually matched control subjects on any demographic variable or on height and weight (table 1). Table 2 presents structure volumes for the autistic and control subjects. Amygdala volume was significantly smaller for the autistic than for control subjects, both with (p = 0.006) and without (p = 0.01) correction for total brain volume. Total brain volume and absolute hippocampal volume did not significantly discriminate between groups. When corrected for total brain volume, hippocampal volume was significantly reduced in the autistic group (p = 0.04). A repeated-measures analysis of variance indicated no significant interaction between side (left versus right) and diagnosis (autism versus control) for either the hippocampus (F = 0.036, p = 0.85) or amygdala (F = 0.003, p = 0.95), indicating that degree of asymmetry did not differ between the two groups.
Structure volumes for autistic subjects and individually matched control subjects
Because of the controversy about brain enlargement in people with autism, it might be argued that height is a better correction factor than brain volume for assessing group differences in specific brain regions. Analyses were repeated using height as a covariate. These analyses are associated with a loss of power, however, because they do not take advantage of the careful individual matching of autistic and control subjects. These analyses yielded significant group differences for amygdala (F = 11.23, p = 0.003) and a trend toward significance for hippocampus (F = 3.59, p = 0.073).
Discussion.
This study provides the first evidence of volumetric abnormalities of the amygdala and hippocampus in vivo in autism. The volume abnormalities were bilateral, with no between-group difference in the symmetry of amygdala and hippocampal volumes. A secondary finding was the absence of a significant increase in total brain volume in these autistic subjects. Because the autistic subjects were not mentally retarded, the reduced volumes of the amygdala and hippocampus are presumed to reflect and be specifically associated with the subjects’ autistic features, rather than a possible reflection of the mental retardation that commonly accompanies autism. The volume reduction was most pronounced for the amygdala, which was significantly smaller in the autistic subjects both with and without consideration of total brain volume. For the hippocampus, the reduction in volume in the autistic subjects was apparent only when hippocampal volume was considered in relation to total brain volume. The more pronounced differences in structure volumes with correction for total brain volume appeared to reflect the effect of slightly greater total brain volumes and the smaller volumes of the amygdala and hippocampus in the autistic subjects.
The reduction in the MRI volumes of the amygdala and hippocampus is consistent with the histopathologic observations by Bauman and Kemper3,9,10 of increased packing density of small, immature-appearing neurons with truncated dendritic development indicative of curtailment in the maturation of the neurons and neuropil. This histopathologic feature likewise suggests that a reduction in the neuropil of the amygdala and hippocampus, and, by inference, their connectivity with other brain regions, and not a destructive lesion, is the most likely basis for the MRI volume reductions in these structures. In a recent preliminary neuropathologic study in which a single slice of hippocampus was examined, Bailey et al.2 did not observe changes in cell packing density. Because Golgi studies were not performed by Bailey et al., dendritic development in hippocampal neurons was not assessed. Furthermore, examination of the amygdala was limited by tissue sampling for neurochemistry.
The more pronounced reduction in amygdala volume and more subtle reduction in hippocampal volume in the non–mentally retarded autistic subjects in this study are consistent with the more extensive histologic abnormalities in the amygdala and less extensive abnormalities in the hippocampus observed by Bauman and Kemper10 in one non–mentally retarded autistic subject, compared with autistic subjects who were mentally retarded. Preliminary neuropathologic examination of the brain in patients with Asperger’s syndrome, a disorder in which the deficits are of the same quality but of lesser severity than in most cases of autism, has revealed similar histologic changes in the amygdala but not the hippocampus.9 These cases of high-functioning autism and Asperger’s syndrome suggest that the amygdala and its neural connections may play a more essential role than the hippocampus in the pathophysiology of the behavioral features of autism.
The findings of the current study are not consistent with the two volumetric imaging studies of hippocampus to date in autism, which have reported negative findings,12,13 even when controlling for total brain volume.13 Differences between our findings and those of previous studies may reflect differences in samples. For example, our sample was higher functioning than previous samples because both Verbal IQ and Full Scale IQ were required to be above 80. In the study by Saitoh et al.,13 Verbal IQ or Performance IQ was required to be above 70, and in the study by Piven et al.,13 Performance IQ was required to be above 70; these IQ criteria typically result in autistic groups that are more severely affected. Methodologic differences between the studies might also be responsible for differences in findings. Saitoh et al.12 performed measurements on only three 5-mm cross-sectional slices through the midsection of the hippocampus, and no attempt was made to match subjects and controls on IQ. In the study by Piven et al.,13 methodology for the volumetric measurements was similar to that used in the current study, although the image analysis program used in the current study allows the rater to trace the borders (not just visualize structures) simultaneously in the three planes. The autistic and control groups in the study by Piven et al.13 were not matched on age, sex, or IQ, variables known to have a significant influence on brain morphology, although these variables were used as covariates at the time of statistical analysis to control for the effect of group differences. The detection of volume differences in the current study may reflect the combined use of true volume measurements, more accurate tracing of margins, and more stringent control of confounding variables by individual matching at subject selection, as well as sample differences.
The amygdala and hippocampus have long been thought to play a role in the pathophysiology of this syndrome, and the findings of the current study provide support for the involvement of these structures. The function of the amygdala has most often been considered in relation to its role in the evaluation of emotional stimuli20 and the impact of emotional intensity, particularly fear, on memory and learning. Recent functional MRI and PET studies21,22 have demonstrated increased activity in the amygdala when normal subjects viewed happy as well as fearful facial expressions compared with emotionally neutral faces. Animal studies have emphasized the impact of early amygdala damage on social behavior in monkeys, with lesions during infancy producing lack of interest and ineptness in social contact as well as lack of facial expression.23 Early neurobehavioral models proposed that the abnormal social behavior in autism was the result of the failure of the amygdala to attach emotional valence to people, reflecting observations of the absence of social behavior and affect in severely autistic people. However, subsequent research demonstrated that the social and affective deficits each were deficits in their own right.24 Investigation of the social deficit demonstrated that a substantial proportion of autistic people develop attachments to people and experience basic emotional states.24 Major deficits were documented in social cognition, particularly in the capacity for making inferences about the beliefs, desires, emotions, and intentions of other people (“theory of mind” deficits).25 These deficits in social cognition are now widely recognized as playing the predominant role in the social manifestations of autism. The cognitive nature of these deficits has suggested that the social behavior is most likely related to the underdevelopment of extensive neural systems that prominently involve cerebral cortex, rather than localized dysfunction of amygdala. This localization is supported by a PET study of brain activation during a “theory of mind” task, reporting that subjects with Asperger’s syndrome did not show the normal activation of the medial part of the left prefrontal area, but instead showed significant activation of a neighboring area, suggesting that they were using a more “general-purpose” reasoning mechanism to infer mental states.26
Investigation of affect and emotion in autism has demonstrated a reduced capacity for exhibiting and comprehending emotions in facial expressions and prosody, for experiencing emotions, and for cognitively defining more abstract emotions.24,27 Less impaired autistic people develop competency with basic emotional states, but lack comprehension and expression abilities for the more subtle emotional shadings and for the cognitive concepts of more abstract emotions.24,27 These higher-order deficits in emotion, affect, and emotional cognition likewise suggest that the neurobiological basis of the emotional and affective deficits in autism involves not only the amygdala but more extensive neural systems and connections with other brain regions, particularly cerebral cortex. The histopathologic features of the amygdala in autism also provide evidence that it is not localized destruction of the amygdala that underlies dysfunction, but underdevelopment of what is likely to be extensive neural connections and neural systems that subserve social and emotional capacities.
The hippocampus is best known for its role in associative memory and in amnesia, and an amnesic disorder was hypothesized as the cause of autism for several decades. However, the neuropsychological evidence supporting the analogy to amnesia was limited to a single study,28 and the memory deficits reported were likely a reflection of the failure to match autistic and control subjects on ability. Multiple subsequent studies of memory using the more specific current diagnostic criteria for autism and screening for other disorders have consistently demonstrated the integrity of associative memory processes, including immediate recall and recognition memory.29,30 Such studies have instead found evidence of reduced use of the cognitive organizing strategies that support memory, deficits more consistent with dysfunction in neocortical memory systems than with localized hippocampal dysfunction. This pattern of preserved associative memory and impaired metamemory function is consistent with the histopathologic evidence indicative of reduced neural connections between hippocampus and other brain regions.
Results of the current study also revealed no significant differences in brain volume between autistic and control subjects. One prior MRI study reported a significant increase in brain volume in autistic subjects of the same age but lower IQ than the subjects in this study.31 Neuropathologic studies2,9,10 have reported increased brain weight in a subgroup of cases, and megaloencephaly has been reported in 2% to 42% of subjects with autism,32 with greater incidence among children than among adults or neonates.33,34 The influence of age and IQ, a commonly used index of autism severity, is unknown. The failure of the current study to document an increase in brain volume may reflect the higher level of functioning among our subjects, the small sample size, or the occurrence of increased brain weight and volume in only a subgroup of autistic people.
The methodologic advantages of this study include the rigorous control of age, IQ, and sex effects through individual matching and prospectively selected volunteers from the community who were screened for both personal and family history of neuropsychiatric disorders. Methodologic improvements in measurement of the hippocampus and amygdala were facilitated by the acquisition of MRI scans with 1.5-mm slice thickness and the capability for viewing and outlining structures simultaneously in multiple imaging planes. The major limitations of the study were the relatively small sample size and the absence of measurements of other structures known to be functionally related to amygdala and hippocampus. Additional imaging studies are needed in autism to establish large sample sizes that document the influence of age and IQ and allow correlations between brain structures thought to be functionally connected to hippocampus and amygdala.
Acknowledgments
Supported by NINDS grant NS 33355 and NICHD grant HD 35469 to Dr. Minshew and by Medical Research Service, Department of Veterans Affairs.
- Received December 21, 1998.
- Accepted July 29, 1999.
References
- ↵
American Psychiatric Association.Diagnostic and statistical manual of mental disorders. 4th ed.Washington, DC:American Psychiatric Association, 1994.
- ↵
Bailey A, Luthert P, Dean A, et al. A clinicopathological study of autism. Brain 1998;121:889–905.
- ↵
- ↵
-
Minshew NJ, Luna B, Sweeney JA. Oculomotor evidence for neocortical systems but not cerebellar dysfunction in autism. Neurology 1999;52:917–922.
- ↵
- ↵
- ↵
- ↵
Bauman M, Kemper T. Neuroanatomic observations of the brain in autism. In: Bauman ML, Kemper TL, eds. The neurobiology of autism. Baltimore:Johns Hopkins University Press, 1994:119–145.
- ↵
- ↵
Saitoh O, Courchesne E, Egaas B, Lincoln AJ, Schreibman L. Cross-sectional area of the posterior hippocampus in autistic patients with cerebellar and corpus callosum abnormalities. Neurology 1995;45:317–324.
- ↵
- ↵
Minshew NJ. Autism. In: Berg BO, ed. Principles of child neurology. New York:McGraw-Hill, 1996:1713–1730.
- ↵
- ↵
- ↵
- ↵
Aylward EH, Anderson NB, Bylsma FW, et al. Frontal lobe volume in patients with Huntington’s disease. Neurology 1998;50:252–258.
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
Baron-Cohen S. Mindblindness: an essay on autism and theory of mind. Cambridge, MA:MIT Press, 1995.
- ↵
- ↵
- ↵
Boucher J, Warrington EK. Memory deficits in early infantile autism: some similarities to the amnesic syndrome. Br J Psychiatry 1976;67:73–87.
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
Mason-Brothers A, Ritvo ER, Pingree C, et al. The UCLA-University of Utah epidemiologic survey of autism: prenatal, perinatal, and postnatal factors. Pediatrics 1990;86:514–519.
Letters: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Hemiplegic Migraine Associated With PRRT2 Variations A Clinical and Genetic Study
Dr. Robert Shapiro and Dr. Amynah Pradhan
Related Articles
- No related articles found.